inprogress

[jalview.git] / forester / java / src / org / forester / ws / seqdb / UniProtEntry.java

diff --git a/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java b/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java
index 4a8d158..00f4b67 100644 (file)
--- a/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java
+++ b/forester/java/src/org/forester/ws/seqdb/UniProtEntry.java
@@ -25,8 +25,9 @@
 
 package org.forester.ws.seqdb;
 
-import java.util.ArrayList;
 import java.util.List;
+import java.util.SortedSet;
+import java.util.TreeSet;
 import java.util.regex.Matcher;
 import java.util.regex.Pattern;
 
@@ -34,6 +35,7 @@ import org.forester.go.BasicGoTerm;
 import org.forester.go.GoNameSpace;
 import org.forester.go.GoTerm;
 import org.forester.phylogeny.data.Accession;
+import org.forester.phylogeny.data.Annotation;
 import org.forester.util.ForesterUtil;
 
 public final class UniProtEntry implements SequenceDatabaseEntry {
@@ -50,9 +52,9 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
     public final static Pattern  PharmGKB_PATTERN  = Pattern.compile( "PharmGKB;\\s+([0-9A-Z]+);" );
     public final static Pattern  Reactome_PATTERN  = Pattern.compile( "Reactome;\\s+([0-9A-Z]+);\\s+([^\\.]+)" );
     private String               _ac;
-    private ArrayList<Accession> _cross_references;
+    private SortedSet<Accession> _cross_references;
     private String               _gene_name;
-    private List<GoTerm>         _go_terms;
+    private SortedSet<GoTerm>    _go_terms;
     private String               _name;
     private String               _os_scientific_name;
     private String               _symbol;
@@ -72,7 +74,7 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
     }
 
     @Override
-    public List<Accession> getCrossReferences() {
+    public SortedSet<Accession> getCrossReferences() {
         return _cross_references;
     }
 
@@ -82,7 +84,7 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
     }
 
     @Override
-    public List<GoTerm> getGoTerms() {
+    public SortedSet<GoTerm> getGoTerms() {
         return _go_terms;
     }
 
@@ -122,15 +124,14 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
 
     private void addCrossReference( final Accession accession ) {
         if ( _cross_references == null ) {
-            _cross_references = new ArrayList<Accession>();
+            _cross_references = new TreeSet<Accession>();
         }
-        System.out.println( "XREF ADDED: " + accession );
         _cross_references.add( accession );
     }
 
     private void addGoTerm( final BasicGoTerm g ) {
         if ( _go_terms == null ) {
-            _go_terms = new ArrayList<GoTerm>();
+            _go_terms = new TreeSet<GoTerm>();
         }
         _go_terms.add( g );
     }
@@ -174,24 +175,24 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
         for( final String line : lines ) {
             //System.out.println( line );
             if ( line.startsWith( "AC" ) ) {
-                e.setAc( DatabaseTools.extract( line, "AC", ";" ) );
+                e.setAc( SequenceDbWsTools.extractFromTo( line, "AC", ";" ) );
             }
             else if ( line.startsWith( "DE" ) && ForesterUtil.isEmpty( e.getSequenceName() ) ) {
                 if ( ( line.indexOf( "RecName:" ) > 0 ) && ( line.indexOf( "Full=" ) > 0 ) ) {
-                    e.setSequenceName( DatabaseTools.extract( line, "Full=", ";" ) );
+                    e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", ";" ) );
                 }
                 else if ( ( line.indexOf( "SubName:" ) > 0 ) && ( line.indexOf( "Full=" ) > 0 ) ) {
-                    e.setSequenceName( DatabaseTools.extract( line, "Full=", ";" ) );
+                    e.setSequenceName( SequenceDbWsTools.extractFromTo( line, "Full=", ";" ) );
                 }
             }
             else if ( line.startsWith( "DE" ) && ForesterUtil.isEmpty( e.getSequenceSymbol() ) ) {
                 if ( line.indexOf( "Short=" ) > 0 ) {
-                    e.setSequenceSymbol( DatabaseTools.extract( line, "Short=", ";" ) );
+                    e.setSequenceSymbol( SequenceDbWsTools.extractFromTo( line, "Short=", ";" ) );
                 }
             }
             else if ( line.startsWith( "GN" ) && ForesterUtil.isEmpty( e.getGeneName() ) ) {
                 if ( line.indexOf( "Name=" ) > 0 ) {
-                    e.setGeneName( DatabaseTools.extract( line, "Name=", ";" ) );
+                    e.setGeneName( SequenceDbWsTools.extractFromTo( line, "Name=", ";" ) );
                 }
             }
             else if ( line.startsWith( "DR" ) ) {
@@ -274,18 +275,33 @@ public final class UniProtEntry implements SequenceDatabaseEntry {
             }
             else if ( line.startsWith( "OS" ) ) {
                 if ( line.indexOf( "(" ) > 0 ) {
-                    e.setOsScientificName( DatabaseTools.extract( line, "OS", "(" ) );
+                    e.setOsScientificName( SequenceDbWsTools.extractFromTo( line, "OS", "(" ) );
                 }
                 else {
-                    e.setOsScientificName( DatabaseTools.extract( line, "OS", "." ) );
+                    e.setOsScientificName( SequenceDbWsTools.extractFromTo( line, "OS", "." ) );
                 }
             }
             else if ( line.startsWith( "OX" ) ) {
                 if ( line.indexOf( "NCBI_TaxID=" ) > 0 ) {
-                    e.setTaxId( DatabaseTools.extract( line, "NCBI_TaxID=", ";" ) );
+                    e.setTaxId( SequenceDbWsTools.extractFromTo( line, "NCBI_TaxID=", ";" ) );
                 }
             }
         }
         return e;
     }
+
+    @Override
+    public SortedSet<Annotation> getAnnotations() {
+        return null;
+    }
+
+    @Override
+    public String getMap() {
+        return null;
+    }
+
+    @Override
+    public String getChromosome() {
+        return null;
+    }
 }